摘要： The joint roughness coefficient (JRC) is an important indicator that characterizes the physical and mechanical behaviors of a jointed rock mass. To investigate the distribution characteristics of the JRC, 31 joint samples belonging to the same joint group with dominant attitude of 209°∠71° were collected from the Guanshan railway tunnel, and the morphologies of the joint samples were digitized using 3D laser scanning. The JRC values in different directions were obtained based on the fractal theory, and the effects of sampling interval on JRC were assessed during the JRC calculation process. The results show that the sampling intervals of profile line and digital point both affect the calculated JRC values. With sampling interval increasing, there exists an obvious threshold above which the JRC value changes from constant to variable. It is found that, the JRC value keeps constant if the profile line sampling interval is smaller than the threshold value (4 mm), which is independent of the roughness degree of the joint. However, the threshold of digital point sampling interval is influenced by joint roughness and has a negative exponential relationship with the JRC value. According to the calculated results of JRC in different directions of the 31 joint samples, it is clear that the JRC values present significant anisotropy and large variation in certain directions. Meanwhile, it is verified that the JRC values in the same direction follow a log-normal distribution. The expected JRC values calculated depend on the probability density function are generally larger than the arithmetic mean values, with a maximum difference up to 23.1%. In consideration of natural distribution, the expected JRC values calculated from probability density function should be more reliable than the arithmetic mean values. Therefore, as a natural property of joint roughness, accurate understanding on the geometrical heterogeneity of JRC will have significant effects on the evaluation of mechanical effects of rock mass structure.